Re: [Discuss-gnuradio] USRP2 Announcement

[Eric Blossom]

On Fri, Sep 12, 2008 at 01:54:08PM -0500, Douglas Geiger wrote:
> Very exciting!
>
> One technical question:
> By 25 MHz of instantaneous RF bandwidth - do you mean the 100Msamples/s 
> from the ADC gets decimated down to 50Msamples/s?  In which case, is the 
> Gig-E able to handle that much sustained throughput (I'm guessing that's 
> with 8-bit samples?).

100MS/s I & Q is decimated to 25MS/s complex.  We use 16-bit I & Q.
That works out to ~800Mbit/s on the gigabit ethernet,  which the USRP2
can sustain, no problem.  

> Also, does the locking of the clocks of multiple USRP2's work like the 
> USRP, e.g. where I can set one as the master and the other as slave - i.e. 
> is the MIMO cable you refer to a coax SMA-SMA cable to connect the clock 
> in/out connect (which have to be soldered in place) and the 2-wire cable to 
> the daughterboard (like described on 
> http://gnuradio.org/trac/wiki/MultiUsrp)?

No soldering required.  The mimo cable is a serial attached scsi cable
repurposed for our needs.  The host configures the clocking on the
USRP2s over the ethernet.

Here are the clocking options:

  USRP uses it's own free running xtal oscillator
  USRP uses the external reference input (SMA connector)
  USRP uses the clock provided over the MIMO cable.

In addition, the master is programmed to drive clock onto the 
MIMO cable.

Eric